LayoutTests/animations/play-state.html fails because of this bug.
Created attachment 50915 [details] Fix for how GraphicsLayer deals with animation states
LGTM (we worked on this together :))
Created attachment 51021 [details] Fix for how GraphicsLayer deals with animation states Updated the patch because it caused a compiler warning.
How can this patch fix a layout test if this code isn't even executed during the DRT run? :) Note that DumpRenderTree doesn't use QGraphicsWebView but QWebView, so the accelerated compositing code path isn't taken AFAICS.
We tested that layout-test manually. It's one of the tests that failed originally when we enabled compositing by default, which raises the question, how did the tests fail in the first place if DRT isn't called? :) in any case, it does fix the layout-test even if it's run manually and not through DRT.
(In reply to comment #5) > We tested that layout-test manually. > It's one of the tests that failed originally when we enabled compositing by > default, which raises the question, how did the tests fail in the first place > if DRT isn't called? :) > > in any case, it does fix the layout-test even if it's run manually and not > through DRT. That is actually a very good point! Strange that enabling AC by default broke 3 tests. I think it is about time we change DRT to use the QGraphicsWebView.
(In reply to comment #5) > We tested that layout-test manually. > It's one of the tests that failed originally when we enabled compositing by > default, which raises the question, how did the tests fail in the first place > if DRT isn't called? :) It sounds like there's a side-effect somewhere... > in any case, it does fix the layout-test even if it's run manually and not > through DRT. Okay, sounds good. Just wanted to check :)
(In reply to comment #6) > (In reply to comment #5) > > We tested that layout-test manually. > > It's one of the tests that failed originally when we enabled compositing by > > default, which raises the question, how did the tests fail in the first place > > if DRT isn't called? :) > > > > in any case, it does fix the layout-test even if it's run manually and not > > through DRT. > > That is actually a very good point! Strange that enabling AC by default broke 3 > tests. > > I think it is about time we change DRT to use the QGraphicsWebView. We should do that at the same time as changing QtLauncher's default then.
Comment on attachment 51021 [details] Fix for how GraphicsLayer deals with animation states Clearing flags on attachment: 51021 Committed r56163: <http://trac.webkit.org/changeset/56163>
All reviewed patches have been landed. Closing bug.